Normalize guessed languages for code highlighting (#36450)

For when Enry correctly recognized the language, but returns the
language name in a way that isn't recognized by chroma.

Resolves https://github.com/go-gitea/gitea/issues/22443

---------

Co-authored-by: Moritz Jörg <moritz.jorg@oceanbox.io>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>

custom/conf/app.example.ini

@@ -2485,8 +2485,9 @@ LEVEL = Info
 ;[highlight.mapping]
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
-;; Extension mapping to highlight class
-;; e.g. .toml=ini
+;; Extension mapping to highlight class, for example:
+;; .toml = ini
+;; .my-js = JavaScript
 
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

modules/analyze/code_language.go

@@ -4,12 +4,13 @@
 package analyze
 
 import (
-	"path/filepath"
+	"path"
 
 	"github.com/go-enry/go-enry/v2"
 )
 
 // GetCodeLanguage detects code language based on file name and content
+// It can be slow when the content is used for detection
 func GetCodeLanguage(filename string, content []byte) string {
 	if language, ok := enry.GetLanguageByExtension(filename); ok {
 		return language
@@ -23,5 +24,5 @@ func GetCodeLanguage(filename string, content []byte) string {
 		return enry.OtherLanguage
 	}
 
-	return enry.GetLanguage(filepath.Base(filename), content)
+	return enry.GetLanguage(path.Base(filename), content)
 }

modules/highlight/highlight.go

@@ -12,7 +12,6 @@ import (
 	"html/template"
 	"io"
 	"path"
-	"path/filepath"
 	"strings"
 	"sync"
 
@@ -25,35 +24,32 @@ import (
 	"github.com/alecthomas/chroma/v2/formatters/html"
 	"github.com/alecthomas/chroma/v2/lexers"
 	"github.com/alecthomas/chroma/v2/styles"
-	lru "github.com/hashicorp/golang-lru/v2"
+	"github.com/go-enry/go-enry/v2"
 )
 
 // don't index files larger than this many bytes for performance purposes
 const sizeLimit = 1024 * 1024
 
+type globalVarsType struct {
+	highlightMapping map[string]string
+	githubStyles     *chroma.Style
+}
+
 var (
-	// For custom user mapping
-	highlightMapping = map[string]string{}
-
-	once sync.Once
-
-	cache *lru.TwoQueueCache[string, any]
-
-	githubStyles = styles.Get("github")
+	globalVarsMu  sync.Mutex
+	globalVarsPtr *globalVarsType
 )
 
-// NewContext loads custom highlight map from local config
-func NewContext() {
-	once.Do(func() {
-		highlightMapping = setting.GetHighlightMapping()
-
-		// The size 512 is simply a conservative rule of thumb
-		c, err := lru.New2Q[string, any](512)
-		if err != nil {
-			panic(fmt.Sprintf("failed to initialize LRU cache for highlighter: %s", err))
-		}
-		cache = c
-	})
+func globalVars() *globalVarsType {
+	// in the future, the globalVars might need to be re-initialized when settings change, so don't use sync.Once here
+	globalVarsMu.Lock()
+	defer globalVarsMu.Unlock()
+	if globalVarsPtr == nil {
+		globalVarsPtr = &globalVarsType{}
+		globalVarsPtr.githubStyles = styles.Get("github")
+		globalVarsPtr.highlightMapping = setting.GetHighlightMapping()
+	}
+	return globalVarsPtr
 }
 
 // UnsafeSplitHighlightedLines splits highlighted code into lines preserving HTML tags
@@ -88,10 +84,56 @@ func UnsafeSplitHighlightedLines(code template.HTML) (ret [][]byte) {
 	}
 }
 
-// Code returns an HTML version of code string with chroma syntax highlighting classes and the matched lexer name
-func Code(fileName, language, code string) (output template.HTML, lexerName string) {
-	NewContext()
+func getChromaLexerByLanguage(fileName, lang string) chroma.Lexer {
+	lang, _, _ = strings.Cut(lang, "?") // maybe, the value from gitattributes might contain `?` parameters?
+	ext := path.Ext(fileName)
+	// the "lang" might come from enry, it has different naming for some languages
+	switch lang {
+	case "F#":
+		lang = "FSharp"
+	case "Pascal":
+		lang = "ObjectPascal"
+	case "C":
+		if ext == ".C" || ext == ".H" {
+			lang = "C++"
+		}
+	}
+	// lexers.Get is slow if the language name can't be matched directly: it does extra "Match" call to iterate all lexers
+	return lexers.Get(lang)
+}
 
+// GetChromaLexerWithFallback returns a chroma lexer by given file name, language and code content. All parameters can be optional.
+// When code content is provided, it will be slow if no lexer is found by file name or language.
+// If no lexer is found, it will return the fallback lexer.
+func GetChromaLexerWithFallback(fileName, lang string, code []byte) (lexer chroma.Lexer) {
+	if lang != "" {
+		lexer = getChromaLexerByLanguage(fileName, lang)
+	}
+
+	if lexer == nil {
+		fileExt := path.Ext(fileName)
+		if val, ok := globalVars().highlightMapping[fileExt]; ok {
+			lexer = getChromaLexerByLanguage(fileName, val) // use mapped value to find lexer
+		}
+	}
+
+	if lexer == nil {
+		// when using "code" to detect, analyze.GetCodeLanguage is slower, it iterates many rules to detect language from content
+		// this is the old logic: use enry to detect language, and use chroma to render, but their naming is different for some languages
+		enryLanguage := analyze.GetCodeLanguage(fileName, code)
+		lexer = getChromaLexerByLanguage(fileName, enryLanguage)
+		if lexer == nil {
+			if enryLanguage != enry.OtherLanguage {
+				log.Warn("No chroma lexer found for enry detected language: %s (file: %s), need to fix the language mapping between enry and chroma.", enryLanguage, fileName)
+			}
+			lexer = lexers.Match(fileName) // lexers.Match will search by its basename and extname
+		}
+	}
+
+	return util.IfZero(lexer, lexers.Fallback)
+}
+
+func renderCode(fileName, language, code string, slowGuess bool) (output template.HTML, lexerName string) {
 	// diff view newline will be passed as empty, change to literal '\n' so it can be copied
 	// preserve literal newline in blame view
 	if code == "" || code == "\n" {
@@ -102,45 +144,25 @@ func Code(fileName, language, code string) (output template.HTML, lexerName stri
 		return template.HTML(template.HTMLEscapeString(code)), ""
 	}
 
-	var lexer chroma.Lexer
-
-	if len(language) > 0 {
-		lexer = lexers.Get(language)
-
-		if lexer == nil {
-			// Attempt stripping off the '?'
-			if before, _, ok := strings.Cut(language, "?"); ok {
-				lexer = lexers.Get(before)
-			}
-		}
+	var codeForGuessLexer []byte
+	if slowGuess {
+		// it is slower to guess lexer by code content, so only do it when necessary
+		codeForGuessLexer = util.UnsafeStringToBytes(code)
 	}
-
-	if lexer == nil {
-		if val, ok := highlightMapping[path.Ext(fileName)]; ok {
-			// use mapped value to find lexer
-			lexer = lexers.Get(val)
-		}
-	}
-
-	if lexer == nil {
-		if l, ok := cache.Get(fileName); ok {
-			lexer = l.(chroma.Lexer)
-		}
-	}
-
-	if lexer == nil {
-		lexer = lexers.Match(fileName)
-		if lexer == nil {
-			lexer = lexers.Fallback
-		}
-		cache.Add(fileName, lexer)
-	}
-
-	return CodeFromLexer(lexer, code), formatLexerName(lexer.Config().Name)
+	lexer := GetChromaLexerWithFallback(fileName, language, codeForGuessLexer)
+	return RenderCodeByLexer(lexer, code), formatLexerName(lexer.Config().Name)
 }
 
-// CodeFromLexer returns a HTML version of code string with chroma syntax highlighting classes
-func CodeFromLexer(lexer chroma.Lexer, code string) template.HTML {
+func RenderCodeFast(fileName, language, code string) (output template.HTML, lexerName string) {
+	return renderCode(fileName, language, code, false)
+}
+
+func RenderCodeSlowGuess(fileName, language, code string) (output template.HTML, lexerName string) {
+	return renderCode(fileName, language, code, true)
+}
+
+// RenderCodeByLexer returns a HTML version of code string with chroma syntax highlighting classes
+func RenderCodeByLexer(lexer chroma.Lexer, code string) template.HTML {
 	formatter := html.New(html.WithClasses(true),
 		html.WithLineNumbers(false),
 		html.PreventSurroundingPre(true),
@@ -155,7 +177,7 @@ func CodeFromLexer(lexer chroma.Lexer, code string) template.HTML {
 		return template.HTML(template.HTMLEscapeString(code))
 	}
 	// style not used for live site but need to pass something
-	err = formatter.Format(htmlw, githubStyles, iterator)
+	err = formatter.Format(htmlw, globalVars().githubStyles, iterator)
 	if err != nil {
 		log.Error("Can't format code: %v", err)
 		return template.HTML(template.HTMLEscapeString(code))
@@ -167,12 +189,10 @@ func CodeFromLexer(lexer chroma.Lexer, code string) template.HTML {
 	return template.HTML(strings.TrimSuffix(htmlbuf.String(), "\n"))
 }
 
-// File returns a slice of chroma syntax highlighted HTML lines of code and the matched lexer name
-func File(fileName, language string, code []byte) ([]template.HTML, string, error) {
-	NewContext()
-
+// RenderFullFile returns a slice of chroma syntax highlighted HTML lines of code and the matched lexer name
+func RenderFullFile(fileName, language string, code []byte) ([]template.HTML, string, error) {
 	if len(code) > sizeLimit {
-		return PlainText(code), "", nil
+		return RenderPlainText(code), "", nil
 	}
 
 	formatter := html.New(html.WithClasses(true),
@@ -180,31 +200,7 @@ func File(fileName, language string, code []byte) ([]template.HTML, string, erro
 		html.PreventSurroundingPre(true),
 	)
 
-	var lexer chroma.Lexer
-
-	// provided language overrides everything
-	if language != "" {
-		lexer = lexers.Get(language)
-	}
-
-	if lexer == nil {
-		if val, ok := highlightMapping[filepath.Ext(fileName)]; ok {
-			lexer = lexers.Get(val)
-		}
-	}
-
-	if lexer == nil {
-		guessLanguage := analyze.GetCodeLanguage(fileName, code)
-
-		lexer = lexers.Get(guessLanguage)
-		if lexer == nil {
-			lexer = lexers.Match(fileName)
-			if lexer == nil {
-				lexer = lexers.Fallback
-			}
-		}
-	}
-
+	lexer := GetChromaLexerWithFallback(fileName, language, code)
 	lexerName := formatLexerName(lexer.Config().Name)
 
 	iterator, err := lexer.Tokenise(nil, string(code))
@@ -218,7 +214,7 @@ func File(fileName, language string, code []byte) ([]template.HTML, string, erro
 	lines := make([]template.HTML, 0, len(tokensLines))
 	for _, tokens := range tokensLines {
 		iterator = chroma.Literator(tokens...)
-		err = formatter.Format(htmlBuf, githubStyles, iterator)
+		err = formatter.Format(htmlBuf, globalVars().githubStyles, iterator)
 		if err != nil {
 			return nil, "", fmt.Errorf("can't format code: %w", err)
 		}
@@ -229,8 +225,8 @@ func File(fileName, language string, code []byte) ([]template.HTML, string, erro
 	return lines, lexerName, nil
 }
 
-// PlainText returns non-highlighted HTML for code
-func PlainText(code []byte) []template.HTML {
+// RenderPlainText returns non-highlighted HTML for code
+func RenderPlainText(code []byte) []template.HTML {
 	r := bufio.NewReader(bytes.NewReader(code))
 	m := make([]template.HTML, 0, bytes.Count(code, []byte{'\n'})+1)
 	for {

modules/highlight/highlight_test.go

@@ -112,7 +112,7 @@ c=2
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			out, lexerName, err := File(tt.name, "", []byte(tt.code))
+			out, lexerName, err := RenderFullFile(tt.name, "", []byte(tt.code))
 			assert.NoError(t, err)
 			assert.Equal(t, tt.want, out)
 			assert.Equal(t, tt.lexerName, lexerName)
@@ -176,7 +176,7 @@ c=2`),
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			out := PlainText([]byte(tt.code))
+			out := RenderPlainText([]byte(tt.code))
 			assert.Equal(t, tt.want, out)
 		})
 	}
@@ -199,3 +199,36 @@ func TestUnsafeSplitHighlightedLines(t *testing.T) {
 	assert.Equal(t, "<span>a</span>\n", string(ret[0]))
 	assert.Equal(t, "<span>b\n</span>", string(ret[1]))
 }
+
+func TestGetChromaLexer(t *testing.T) {
+	globalVars().highlightMapping[".my-html"] = "HTML"
+	t.Cleanup(func() { delete(globalVars().highlightMapping, ".my-html") })
+
+	cases := []struct {
+		fileName string
+		language string
+		content  string
+		expected string
+	}{
+		{"test.py", "", "", "Python"},
+
+		{"any-file", "javascript", "", "JavaScript"},
+		{"any-file", "", "/* vim: set filetype=python */", "Python"},
+		{"any-file", "", "", "fallback"},
+
+		{"test.fs", "", "", "Forth"},
+		{"test.fs", "F#", "", "FSharp"},
+		{"test.fs", "", "let x = 1", "FSharp"},
+
+		{"test.c", "", "", "C"},
+		{"test.C", "", "", "C++"},
+		{"OLD-CODE.PAS", "", "", "ObjectPascal"},
+		{"test.my-html", "", "", "HTML"},
+	}
+	for _, c := range cases {
+		lexer := GetChromaLexerWithFallback(c.fileName, c.language, []byte(c.content))
+		if assert.NotNil(t, lexer, "case: %+v", c) {
+			assert.Equal(t, c.expected, lexer.Config().Name, "case: %+v", c)
+		}
+	}
+}

modules/indexer/code/search.go

@@ -72,10 +72,10 @@ func writeStrings(buf *bytes.Buffer, strs ...string) error {
 
 func HighlightSearchResultCode(filename, language string, lineNums []int, code string) []*ResultLine {
 	// we should highlight the whole code block first, otherwise it doesn't work well with multiple line highlighting
-	hl, _ := highlight.Code(filename, language, code)
+	hl, _ := highlight.RenderCodeFast(filename, language, code)
 	highlightedLines := strings.Split(string(hl), "\n")
 
-	// The lineNums outputted by highlight.Code might not match the original lineNums, because "highlight" removes the last `\n`
+	// The lineNums outputted by render might not match the original lineNums, because "highlight" removes the last `\n`
 	lines := make([]*ResultLine, min(len(highlightedLines), len(lineNums)))
 	for i := range lines {
 		lines[i] = &ResultLine{

modules/markup/orgmode/orgmode.go

@@ -5,7 +5,6 @@ package orgmode
 
 import (
 	"fmt"
-	"html"
 	"html/template"
 	"io"
 	"strings"
@@ -17,7 +16,6 @@ import (
 	"code.gitea.io/gitea/modules/setting"
 
 	"github.com/alecthomas/chroma/v2"
-	"github.com/alecthomas/chroma/v2/lexers"
 	"github.com/niklasfasching/go-org/org"
 )
 
@@ -57,40 +55,20 @@ func Render(ctx *markup.RenderContext, input io.Reader, output io.Writer) error
 	htmlWriter.HighlightCodeBlock = func(source, lang string, inline bool, params map[string]string) string {
 		defer func() {
 			if err := recover(); err != nil {
+				// catch the panic, log the error and return empty result
 				log.Error("Panic in HighlightCodeBlock: %v\n%s", err, log.Stack(2))
-				panic(err)
 			}
 		}()
-		w := &strings.Builder{}
 
-		lexer := lexers.Get(lang)
-		if lexer == nil && lang == "" {
-			lexer = lexers.Analyse(source)
-			if lexer == nil {
-				lexer = lexers.Fallback
-			}
-			lang = strings.ToLower(lexer.Config().Name)
-		}
+		lexer := highlight.GetChromaLexerWithFallback("", lang, nil) // don't use content to detect, it is too slow
+		lexer = chroma.Coalesce(lexer)
 
+		sb := &strings.Builder{}
 		// include language-x class as part of commonmark spec
-		if err := ctx.RenderInternal.FormatWithSafeAttrs(w, `<pre><code class="chroma language-%s">`, lang); err != nil {
-			return ""
-		}
-		if lexer == nil {
-			if _, err := w.WriteString(html.EscapeString(source)); err != nil {
-				return ""
-			}
-		} else {
-			lexer = chroma.Coalesce(lexer)
-			if _, err := w.WriteString(string(highlight.CodeFromLexer(lexer, source))); err != nil {
-				return ""
-			}
-		}
-		if _, err := w.WriteString("</code></pre>"); err != nil {
-			return ""
-		}
-
-		return w.String()
+		_ = ctx.RenderInternal.FormatWithSafeAttrs(sb, `<pre><code class="chroma language-%s">`, strings.ToLower(lexer.Config().Name))
+		_, _ = sb.WriteString(string(highlight.RenderCodeByLexer(lexer, source)))
+		_, _ = sb.WriteString("</code></pre>")
+		return sb.String()
 	}
 
 	w := &orgWriter{rctx: ctx, HTMLWriter: htmlWriter}

routers/init.go

@@ -15,7 +15,6 @@ import (
 	"code.gitea.io/gitea/modules/eventsource"
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/git/gitcmd"
-	"code.gitea.io/gitea/modules/highlight"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/markup"
 	"code.gitea.io/gitea/modules/markup/external"
@@ -131,7 +130,6 @@ func InitWebInstalled(ctx context.Context) {
 	mustInit(uinotification.Init)
 	mustInitCtx(ctx, archiver.Init)
 
-	highlight.NewContext()
 	external.RegisterRenderers()
 	markup.Init(markup_service.FormalRenderHelperFuncs())
 

routers/web/repo/blame.go

@@ -267,7 +267,7 @@ func renderBlame(ctx *context.Context, blameParts []*gitrepo.BlamePart, commitNa
 
 	bufContent := buf.Bytes()
 	bufContent = charset.ToUTF8(bufContent, charset.ConvertOpts{})
-	highlighted, lexerName := highlight.Code(path.Base(ctx.Repo.TreePath), language, util.UnsafeBytesToString(bufContent))
+	highlighted, lexerName := highlight.RenderCodeSlowGuess(path.Base(ctx.Repo.TreePath), language, util.UnsafeBytesToString(bufContent))
 	unsafeLines := highlight.UnsafeSplitHighlightedLines(highlighted)
 	for i, br := range rows {
 		var line template.HTML

routers/web/repo/view_file.go

@@ -124,11 +124,11 @@ func handleFileViewRenderSource(ctx *context.Context, filename string, attrs *at
 	}
 
 	language := attrs.GetLanguage().Value()
-	fileContent, lexerName, err := highlight.File(filename, language, buf)
+	fileContent, lexerName, err := highlight.RenderFullFile(filename, language, buf)
 	ctx.Data["LexerName"] = lexerName
 	if err != nil {
-		log.Error("highlight.File failed, fallback to plain text: %v", err)
-		fileContent = highlight.PlainText(buf)
+		log.Error("highlight.RenderFullFile failed, fallback to plain text: %v", err)
+		fileContent = highlight.RenderPlainText(buf)
 	}
 	status := &charset.EscapeStatus{}
 	statuses := make([]*charset.EscapeStatus, len(fileContent))

services/gitdiff/gitdiff.go

@@ -331,7 +331,7 @@ func (diffSection *DiffSection) getLineContentForRender(lineIdx int, diffLine *D
 	if setting.Git.DisableDiffHighlight {
 		return template.HTML(html.EscapeString(diffLine.Content[1:]))
 	}
-	h, _ = highlight.Code(diffSection.FileName, fileLanguage, diffLine.Content[1:])
+	h, _ = highlight.RenderCodeFast(diffSection.FileName, fileLanguage, diffLine.Content[1:])
 	return h
 }
 
@@ -1349,7 +1349,7 @@ func GetDiffForRender(ctx context.Context, repoLink string, gitRepo *git.Reposit
 
 func highlightCodeLines(diffFile *DiffFile, isLeft bool, rawContent []byte) map[int]template.HTML {
 	content := util.UnsafeBytesToString(charset.ToUTF8(rawContent, charset.ConvertOpts{}))
-	highlightedNewContent, _ := highlight.Code(diffFile.Name, diffFile.Language, content)
+	highlightedNewContent, _ := highlight.RenderCodeFast(diffFile.Name, diffFile.Language, content)
 	unsafeLines := highlight.UnsafeSplitHighlightedLines(highlightedNewContent)
 	lines := make(map[int]template.HTML, len(unsafeLines))
 	// only save the highlighted lines we need, but not the whole file, to save memory